Models & Labs
Claude 4.6 and New Model Releases Announced
Lev Selectorhigh confidence
Why it matters
- →The introduction of these models signifies continuous improvement in AI capabilities, offering more options for developers.
Claude 4.6 has been released, showcasing its capabilities in simulated business environments. Alongside this, new models such as Mimo v2 with 15 billion parameters, MiniMax 2.5, GLM-5 from Z.ai, and GPT-5.3-Codex-Spark have also been introduced. These advancements reflect the ongoing evolution of AI models, enhancing their reasoning and operational capabilities. The release of these models is expected to provide developers with more powerful tools for various applications.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
More in Models & Labs
Models & Labsother
llama.cpp b12340 adds CUDA 13 and ROCm 10.0
This release quietly closes the hardware gap for local inference by adding default support for CUDA 13 and ROCm 10.0 across Linux and Windows. Users with newer NVIDIA GPUs or AMD accelerators no longer need to compile from source to get performance, while Apple Silicon builds now ship KleidiAI disabled by default, shifting the focus back to standard ARM optimizations. The inclusion of Snapdragon support for Linux and Android marks a significant step toward on-device AI beyond the x86/ARM desktop ecosystem. It is less about new features and more about ensuring llama.cpp remains the universal runtime for every available accelerator.
llama.cpp Releases
Models & Labscoding
llama.cpp b12342 adds few-row MMQ for speculative decoding
Speculative decoding just got a significant speed boost on NVIDIA GPUs. This release introduces a specialized kernel for 'few-row' matrix multiplications, targeting the small batch sizes typical of verification passes in speculative decoding workflows. By optimizing these specific operations for quantized formats like Q8_0 and IQ4 variants, llama.cpp reduces latency where it matters most during token generation. The change is CUDA-only for now but leverages hardware capabilities from Ampere onwards, making local inference faster without requiring new model architectures or complex configuration changes.
